This paper attempts to investigate trade potential for Korea and each of EU 25 Member Countries using modified gravity model. To avoid biased results stemming from simple OLS estimation, especially heteroskedasticity problem, various estimation techniques were used, i.e., fixed effect, random effect, between effect, Hausman-Taylor, Hausman-Taylor AR(1) and Random Effects GLS with Common AR(1). Considering the magnitude of our regression coefficients, Hausman-Taylor estimation and Random Effects GLS with Common AR(1) seems to be the most relevant estimation methods to our model. The estimated coefficients from the Hausman-Taylor estimation and Random Effects GLS with Common AR(1) were then employed to predict those bilateral trade potentials. The results show that Korea exported to existing EU Members more than the predicted potential trade level in 2006. Compared with those in 2000, this was even more in terms of absolute export value and the ratio of actual export value to potential export value. In addition, Korea exported more to 5 newly acceding CEEC countries in 2006. Consequently, it does not seem to have been any trade diversion effect vis-a-vis Korea.